Factors Predictive of Survival in Advanced Laryngeal Cancer

Objective  To determine the factors predictive of improved survival among patients with advanced laryngeal cancer.

Design  National hospital-based cancer registry.

Patients  Patients treated with total laryngectomy (TL), radiation therapy alone (RT), or combined chemotherapy and radiation therapy (chemo-RT).

Results  Of the 10 590 patients meeting the initial inclusion criteria, 7019 had appropriate nonmissing values in all study variables and were included in the analysis. Overall, TL was significantly associated with increased likelihood of survival compared with RT or chemo-RT (P<.001). Among patients with stage III cancer, TL and chemo-RT had similar impacts on survival (each showed increased survival compared with RT), whereas TL was associated with significantly greater survival than chemo-RT or RT among patients with stage IV disease (P<.001). Overall survival was also decreased among men, black patients (compared with white patients), and patients with Medicare or Medicaid or those who were uninsured (compared with those with private insurance).

Conclusions  Among patients with the most advanced disease (stage IV), TL was associated with increased survival compared with chemo-RT or RT, whereas both TL and chemo-RT improved survival over RT among patients with stage III cancer. Insurance type and black race also showed significant associations with survival, which may reflect barriers in access to care (P<.001).

Laryngeal cancer is diagnosed in nearly 10 000 men and women in the United States annually and is among the most common cancers of the upper aerodigestive tract.¹ For many years, total laryngectomy (TL) followed by radiotherapy (RT) was considered the standard treatment for advanced laryngeal cancer. Unfortunately, patients treated with TL experience a complete loss of voice and may also experience impairment of swallowing function, leading to decreased quality of life in many aspects, including nutrition, social functioning, and personal hygiene.² In 1991, results of a prospective, randomized clinical trial, the Veterans Affairs (VA) Laryngeal Cancer Study,³ were published. That study supported the use of induction chemotherapy followed by RT for patients with advanced laryngeal cancer who wished for laryngeal-conserving therapy. In this landmark study, more than 300 patients with stage III or stage IV laryngeal cancer were randomized to 2 arms of treatment: induction chemotherapy with cisplatin and fluorouracil followed by RT or TL followed by RT. In most patients in the chemotherapy and RT arm (62%), the larynx was preserved. Furthermore, no differences in overall survival were evident between the nonsurgical and surgical arms of the study. A similarly designed study in Europe also demonstrated no difference in overall survival and a 42% laryngeal preservation rate.⁴

Following the completion of the VA Laryngeal Cancer Study, the Head and Neck Intergroup in the United States initiated the Radiation Therapy Oncology Group (RTOG) 91-11 study⁵ to determine the precise role of chemotherapy for patients with advanced laryngeal cancer. This 3-arm randomized trial compared RT alone, chemotherapy (cisplatin) followed by RT, and concurrent chemotherapy (cisplatin) with RT. Concurrent chemotherapy with RT (hereinafter, chemo-RT) offered the highest laryngeal preservation rate (84%), which was statistically significant compared with RT alone (70%; P < .001) or induction chemotherapy followed by RT (74%; P = .005). This study solidified the use of chemo-RT as a valid treatment option for patients with advanced laryngeal cancer who desired laryngeal preservation. Because chemo-RT is more toxic, RT alone may still be offered to patients who are infirm or who have limited psychosocial support.

We investigated the clinical and demographic factors that are associated with improved survival among patients with advanced-stage laryngeal cancer, using data collected by the National Cancer Database (NCDB)¹⁴ for patients diagnosed from 1995 to 1998. Our objectives were to determine (1) the relationship between treatment type and survival and (2) to control for potential confounders, including clinical and demographic factors that may have an impact on survival.

The NCDB¹⁴ is a national hospital-based cancer registry jointly sponsored by the American College of Surgeons and the American Cancer Society. Prior to 1997, submission of medical records of patients with cancer to the NCDB was voluntary and open to all cancer facilities in the United States. Beginning in 1997, data collection was restricted to facilities with Commission on Cancer (CoC) approved programs, which were required to report. In recent years, approximately 75% of newly diagnosed cancers in the United States have been captured in the database.⁶ The NCDB¹⁴ contains standardized data elements on patient demographics; tumor characteristics, including stage and histopathologic findings; and the first course of treatment in common with population-based registries. In addition, the NCDB¹⁴ contains information on patient insurance status and county of residence (which is useful for area-based socioeconomic characteristics).

Data reported to the NCDB¹⁴ are retrospective in nature, and no patient or physician identifiers are collected. Case identification information (facility identification number and local registry accession number) is collected for administrative purposes only. Analyses are reported only at the aggregate level to assist hospital cancer programs with quality assurance rather than used to make decisions about individuals and their care. Because no patient, physician, or hospital identifiers are included in the research data set and no protected health information is present, institutional review board approval was not required for this study.

All cases of squamous cell cancer of the larynx diagnosed from 1995 to 1998 were selected from the NCDB.¹⁴ Data were collected according to the Facility Oncology Registry Data Standards,⁷ the American Joint Committee on Cancer's Cancer Staging Manual,⁸ and the second and third editions of the World Health Organization International Classification of Disease for Oncology (ICD-O-2⁹ and ICD-O-3,¹⁰ respectively). The ICD-O-2⁹ codes were recoded to ICD-O-3.¹⁰ Cases of invasive squamous cell carcinoma of the larynx were extracted from the NCDB¹⁴ using the appropriate ICD-O-3 site and histology codes (n = 76 529). Site codes selected were C10.1 (anterior surface of epiglottis), C13.1 (aryepiglottic fold), C32.0 to C32.3 (glottis, supraglottis, subglottis, laryngeal cartilage), C32.8 (overlapping lesion of larynx), and C32.9 (larynx, not otherwise specified [NOS]). Histology codes captured included 8045, 8051, 8052, 8070 to 8078, 8083, 8084, 8090 to 8094, 8097, 8123, 8147, 8560, and 8570 for squamous cell and/or basal squamous cell cancer.

Only adults 18 years or older with advanced laryngeal cancer (stage III and stage IV) who were treated at the reporting institution and whose initial treatment was TL, RT, or chemo-RT were included (n = 10 590). The type of initial treatment referred to the first course of cancer-directed therapy. Those treated with salvage laryngectomy that was performed after other treatments would not be included in this group. We defined the treatment as chemo-RT if the 2 treatments were given concurrently or if the RT start date was within 90 days after the chemotherapy start date; chemo-RT included both induction chemotherapy and concurrent chemoradiation. Radiation therapy was defined as radiation alone if there was no chemotherapy treatment given concurrently or if the date RT started preceded the chemotherapy start date.

The independent variables included in this analysis were divided into 3 categories based on the source or type of information: patient-level demographics, clinical characteristics, and area-level information. Patient-level demographic variables were age at diagnosis, sex, race, and primary payer or insurance type at diagnosis. Clinical characteristics, also captured at the patient level, were overall tumor stage, T category, year of diagnosis, and treatment facility type. Area-level characteristics were based on the patient's residence and consisted of the percentage of high school graduates and median household income in the patient's zip code of residence. We excluded all patients with missing data for the predictor and dependent variables, leaving 7019 patients eligible for the study population.

Race was categorized as white, black, Hispanic, and others (which included Asian American, American Indian, and other), whereas sex was categorized as male and female. Age at diagnosis was a continuous variable for purposes of analysis. Primary payer or insurance type at diagnosis was determined using Facility Oncology Registry Data Standards (FORDS)⁷ codes, which were grouped into the following categories: Medicaid, Medicare (which included both Medicare alone and Medicare with supplement), uninsured (which included FORDS codes for not insured– NOS; not insured–charity write-off, and not insured–self-pay), other government funded plans (VA, Indian Health Service, Public Health Service, welfare, state-funded–NOS, and federally funded–NOS), and private insurance plans (health maintenance organization, preferred provider organization, managed care–NOS, private insurance, Champus or Tricare, military, and insured–NOS). The plans in the private insurance category were grouped together because these plans represent either privately purchased insurance (purchased by the individual, a family member, and/or employer) or insurance provided by the military that functions in a similar manner as private insurance (Champus or Tricare). Because Medicare eligibility for individuals younger than 65 years differs from eligibility for those 65 years or older, the Medicare category was dichotomized for analyses into Medicare among patients 18 to 64 years of age and Medicare among patients 65 years or older.

The overall tumor stage (stage III vs stage IV) and primary T category (T1, T2, T3, or T4) were assessed by clinical staging data. If the clinical stage was not present, the pathological stage was used. The year of diagnosis was analyzed as a continuous variable (1995-1998) for the regression analyses. Data on the 3 types of treatment facilities were extracted from the NCDB¹⁴ CoC hospital category variable. Community hospitals treat at least 300 patients with cancer a year and have a full range of services for cancer care, but patients need referrals for portions of their treatment. Community cancer centers are facilities that offer the same range of services as the community hospitals but treat at least 750 cancer cases annually and conduct weekly cancer conferences. Teaching and/or research facilities differ from community cancer facilities in that the teaching and research facilities have residency programs and ongoing cancer research. Twenty-nine of the 39 National Cancer Institute–designated comprehensive cancer programs participate in the CoC approvals program and are included among teaching and research facilities in this study.

Area-based indicators of patient socioeconomic status—specifically, education and income—were derived from 2000 US census data at the zip code level and shown as quartiles based on the observed distribution in the US population. The percentages of the population without a high school diploma in a patient's zip code of residence are shown as 29.0% or greater, 20.0% to 28.9%, 14.0% to 19.9%, and less than 14.0%; and the median household income is reported as less than $30 000, $30 000 to $34 999, $35 000 to $45 999, and $46 000 or greater.

Analyses were performed with SAS statistical software (version 9.1; SAS Statistical Institute, Cary, North Carolina). Bivariate χ² analysis was used to test for associations among treatment types (TL, chemo-RT, RT) and dichotomous or ordinal-dependent variables included in the analysis, whereas t tests were used to compare continuous variables by treatment type. The impact of treatment group on survival was evaluated using proportional hazards (PH) regression. The number of days from diagnosis until the last contact were used as the time-dependent variable, whereas the vital status at last contact (alive vs dead) was used as the censoring variable (ie, patients alive at last contact were censored). Because follow-up data beyond 5 years after diagnosis are not routinely collected by hospitals reporting to the NCDB,¹⁴ follow-up time was truncated at 5 years, and any patient with follow-up of more than 5 years was classified as censored. Cause-specific mortality was not available in the NCDB. All PH analyses were stratified by age and T category because neither of these variables met the criteria for the PH assumption. The PH analyses were performed for all patients as well as separately for patients with overall stage III disease vs those with stage IV disease, and for those with T3 primary disease. Statistical significance was defined as α <.01, whereas marginal significance was defined as .01 < α <.05.

Table 1 presents the characteristics of each treatment group. Significant differences exist among the 3 groups (P<.001). Of the 7019 patients included in the analysis, 53.6% received TL (hereinafter, the TL group), 30.6% received RT alone (hereinafter, the RT group), and the remaining 15.8% received chemo-RT (hereinafter, the chemo-RT group). Those in the TL group were slightly older than those in the chemo-RT group and slightly younger than those in the chemo-RT group. Those in the TL group were more likely to be male, have stage IV disease, and have longer follow-up times than patients in the other groups. Those in the chemo-RT group were more likely to come from areas of residences (zip codes) with higher percentages of high school graduates and higher household incomes than were patients in either of the other treatment groups, whereas those in the RT group were more likely to receive treatment at community hospitals and less likely to receive treatment at teaching and research hospitals. Those in the TL group were more likely to have T4 disease than were patients in either of the other groups, whereas those in the chemo-RT group were more likely to be diagnosed later in the study period (1998) than were those in the other groups. Compared with the TL group, a significantly greater proportion of the RT group had Medicare for individuals 65 years or older as their primary payer (P<.001); the chemo-RT group had higher rates of private insurance coverage and lower rates of coverage with Medicare for individuals 65 years or older compared with the TL group.

Table 2 presents results from the PH analysis evaluating the impact of treatment group and patient characteristics on survival for all patients with advanced laryngeal cancer. Controlling for the other included factors, the RT and chemo-RT groups had lower odds of survival than did the TL group. The increased risk (hazard ratio) associated with death is approximately 30% for the chemo-RT group and 60% for the RT group. Men and patients with stage IV disease also had decreased odds of survival compared with women and patients with stage III disease, respectively. Black individuals had increased odds of dying compared with white individuals. Zip code–based education and income levels had no notable impacts on mortality, whereas hospital type had only marginal effects.

Uninsured patients and those with Medicaid, Medicare, or other government health plan coverage had a significantly increased risk of mortality compared with those with private insurance (P<.001), with the greatest increased risk observed among Medicaid patients (57%). Although this may represent differential medical care associated with type of health insurance, it more likely relates to multiple socioeconomic status factors correlated with health insurance status. For example, uninsured or underinsured patients are less likely to have usual sources of medical care or to engage in a wide range of preventive care behaviors; these factors may be associated with an increased risk of overall mortality.

Table 3 presents results from separate PH analyses for patients with stage III and stage IV disease. Among patients with stage III disease, controlling for other factors included in the analysis, patients who received RT only had a significantly decreased survival rate compared with those who received TL (P<.001). In contrast, the odds of mortality were not significantly different between patients with stage III disease receiving chemo-RT vs those receiving TL (P = .09). Patients with stage III disease with Medicaid or older patients with Medicare coverage also had an increased likelihood of death compared with those with private insurance, whereas the impact of being uninsured or having other government insurance was marginal. Neither sex (P = .17) nor race or ethnicity (P = .88) was significantly associated with the odds of mortality in the study population. Among patients with stage IV disease, both chemo-RT and RT only were associated with an increased risk of mortality compared with those who received TL. Males also had increased odds of mortality compared with females, and all insurance groups except Medicare for individuals 65 years or older showed increased odds of mortality compared with private insurance.

Because the choice between chemo-RT and TL as optimal treatment for patients with T3 primary cancers is a matter of debate, we also performed separate PH analysis for patients with T3 primary laryngeal cancers (data not shown). Among patients with T3 laryngeal cancer, those treated with chemo-RT had a significantly increased risk for death compared with those treated with TL (hazard ratio, 1.18; P = .03); however, patients with T3 laryngeal cancer treated with RT alone had a markedly increased risk for death compared with those who received TL (hazard ratio = 1.59, P < .001).

Our analysis demonstrates that for patients with advanced laryngeal cancer, several factors have an impact on the likelihood of survival, including type of treatment, sex, race, and insurance status. Our finding of increased risk for death among the RT group (compared with the TL group) is consistent with previous reports.⁵ However, contrary to reports in the literature³ of an equivalent survival rate for those treated with chemo-RT and TL among patients with advanced laryngeal cancer, we found an increased risk of death for patients treated with chemo-RT in the overall study population. Our findings may reflect differences in patient mix between those enrolled in previous clinical trials (reporting equivalent survival following TL and chemo-RT) and the broader population of patients with laryngeal cancer treated at CoC facilities across the United States. We included patients in all practice settings, not only at tertiary care centers where most phase III clinical trials take place. Furthermore, in separate analyses by stage at diagnosis, the difference in likelihood of survival between the TL group and the chemo-RT group was statistically significant only for those with stage IV disease (P <.001). Our findings suggest that physicians should exercise caution in interpreting clinical trial results for use in the broader population, particularly among patients with stage III laryngeal cancer. Although our results demonstrate nearly equivalent efficacy for chemo-RT and TL for T3 and stage III disease, we show significant differences in survival for all patients with advanced laryngeal cancer (P <.001), thus indicating that much of the difference is attributable to stage IV disease and not to stage III disease. The treatment decision (chemo-RT vs TL) for stage III and T3 disease should be based on other factors, such as patient preference, performance status, and function, rather than on survival, especially because there does not seem to be notable differences in this regard.

Differences in outcome by treatment facility as part of this multivariate analysis indicate that the type of facility where patients seek treatment has only a marginal impact on survival. Patients treated at community hospitals had only a marginally greater likelihood of death compared with those treated at teaching and research facilities, whereas there were no notable differences between patients treated at community cancer centers and those treated at teaching and research facilities. When evaluating patients with stage III disease vs those with stage IV disease at diagnosis separately, we found no notable difference associated with facility type.

Compared with white patients, black patients had a decreased likelihood of survival after controlling for clinical factors. This is in agreement with a report¹² examining approximately 100 patients with head and neck carcinoma treated at a university medical center; after controlling for treatment and other clinical factors, the likelihood of death was significantly greater for black patients than for white patients (OR, 1.14; P <.001). In addition, individuals with Medicaid insurance had the highest likelihood of death compared with those with private insurance, although all individuals without private insurance experience increased likelihood of death compared with those with private insurance. We do not believe that insurance status in this analysis represents differential treatment or quality of care for patients with advanced laryngeal cancer. Rather, insurance status is likely a proxy for multiple medical care issues, including usual source of medical care, participation in screening and preventive care activities, and exposure to related risk factors, including alcohol and/or tobacco use and poor diet, all of which can influence overall survival. A recent Canadian study¹³ also reported a difference in cause-specific survival for patients with laryngeal cancer based on socioeconomic status. Our findings indicate that substantial disparities in survival, potentially related to barriers to adequate access to care, exist among US patients with advanced laryngeal cancer. Specifically, patients with identical tumor stage, demographic characteristics, and type of treatment have notable differences in survival related to nonclinical characteristics. Further research is needed to identify these factors influencing survival and evaluate how best to address them.

In conclusion, this analysis demonstrates that TL yields the highest likelihood of survival for patients with advanced laryngeal cancer. In addition, individuals who are white, male, or have private health insurance also have increased likelihood of survival. These results differ from those of previous analyses comparing TL and chemo-RT, suggesting that caution is needed when applying clinical trial findings to broader medical care settings and populations. The different results from an analysis of all practice settings suggest that patient selection and heterogeneity are different from those included in clinical trials.

Correspondence: Amy Y. Chen, MD, MPH, 250 Williams St NW, Atlanta, GA 30303-1002 (amy.chen@emoryhealthcare.org).

Submitted for Publication: April 24, 2007; final revision received July 2, 2007; accepted July 30, 2007.

Author Contributions: Drs Chen and Halpern had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Chen and Halpern. Acquisition of data: Chen. Analysis and interpretation of data: Chen and Halpern. Drafting of the manuscript: Chen and Halpern. Critical revision of the manuscript for important intellectual content: Chen and Halpern. Statistical analysis: Chen and Halpern. Administrative, technical, and material support: Chen. Study supervision: Chen.

Financial Disclosure: None reported.

Previous Presentation: This article was presented at The American Head & Neck Society 2007 Annual Meeting; April 28, 2007; San Diego, California.